Introduction

This feature is experimental and not ready for use in production. It is only available as part of an Early Access Program, and can go under breaking changes until general availability.

Graph Data Science for BigQuery enables data scientists and analysts to analyze data in BigQuery tables using Neo4j Graph Data Science algorithms and write results back to BigQuery tables.

Architecture

Connector architecture
Figure 1. Connector architecture
  1. Data flows into BigQuery, for example via Pub/Sub.

  2. The bigquery-connector Docker image is deployed into the GCP project, where SQL procedures are used to project the graph into an Aura instance.

  3. Analysts can run one of 50+ algorithms (for example FastRP) in Google Colab notebooks.

  4. Results can be written back to BigQuery using the SQL procedures in bigquery-connector.

  5. Analysts and data scientists can continuously refine the algorithms they use or use more advanced techniques.

Supported data types

The connector supports creating in-memory graph projections directly from BigQuery datasets in both AuraDS and Neo4j with GDS. This is the fastest way to ingest data with only the following data types supported.

  • LONG

  • DOUBLE

  • ARRAY<LONG>

  • ARRAY<FLOAT>

  • ARRAY<DOUBLE>